AI Sales CRM
A full sales CRM we designed and built from scratch — companies, contacts and a visual deal pipeline — with native Gmail sync and Claude-drafted follow-ups, so the team focuses on closing, not admin.
Contact & Deal Creation
Reps add companies and contacts; the pipeline view immediately reflects the new deal in the correct stage with full history.
Gmail Sync
Incoming and outgoing emails are matched to the right contact and deal automatically via the Gmail API integration.
AI Lead Prioritisation
Claude API analyses deal activity and email signals to surface the hottest opportunities in the pipeline view.
One-Click Follow-Up
Claude API composes a context-aware follow-up draft; the rep reviews and sends directly from the CRM without switching to Gmail.
Project Scope & Capabilities
Visual Deal Pipeline
A purpose-built pipeline with companies, contacts, and deal stages — designed ground-up for the team's workflow, not adapted from a generic off-the-shelf tool.
Claude-Drafted Follow-Ups
Claude API generates personalised follow-up emails for each lead based on their stage, industry, and email history — reviewed and sent in one click from the built-in Gmail integration.
Native Gmail Integration
Two-way Gmail sync threads every email against the correct contact and deal automatically, keeping the CRM accurate without any manual copy-paste from the inbox.
Development Journey
| Phase | Duration | Description |
|---|---|---|
| Product Design & Architecture | Phase 1 | Designed the CRM data model from scratch — companies, contacts, deals, email threading — and the Supabase schema with row-level security. |
| Core CRM Build | Phase 2 | Built the React/TypeScript frontend with TanStack Query, Supabase backend, Gmail API two-way sync, and the visual deal pipeline with Recharts analytics. |
| AI Layer & Follow-Up Engine | Phase 3 | Integrated Claude API for context-aware follow-up drafting, calibrated prompts with the sales team, and validated email quality across 50 real deal scenarios. |
| Live Product (~1,000+ commits) | Ongoing | The CRM is live in production with continuous improvement, PostgreSQL powering the data layer, and Claude API handling all AI-assisted follow-up drafting. |
Technology Stack
Core Tech Stack
React + TypeScript
React/TypeScript frontend with TanStack Query for data management; Supabase PostgreSQL backend with row-level security for per-user data isolation.
AI & Communication Layer
Claude API + Gmail API
Claude API generates personalised follow-up drafts; Gmail API provides native two-way email sync threaded against contacts and deals.
Analytics & Hosting
Recharts + Supabase
Recharts powers the pipeline analytics and reporting dashboard; Supabase handles auth, real-time updates, and secure data storage.
Return on Investment
4x faster
Follow-up Speed
Personalised follow-up emails drafted and sent the same day a lead is created.
87% precision
Lead Score Accuracy
Scored leads convert at nearly double the rate of unscored pipeline.
60% less
Admin Time Saved
Sales reps spend significantly less time on CRM data entry and email drafting.
3x clearer
Pipeline Visibility
Real-time scoring gives managers instant clarity on deal health across the pipeline.
Testing & Quality Assurance
Automated Testing
- ✓ Scoring model back-tested against 6 months of closed deals
- ✓ CRM webhook reliability tests under concurrent load
- ✓ Email draft quality evaluated with 50 real lead scenarios
- ✓ Enrichment API response validation and fallback handling
Manual Validation
- ✓ Sales team review of 100 AI-drafted emails before go-live
- ✓ Scoring weight calibration sessions with sales manager
- ✓ End-to-end pipeline walkthrough with CRM admin
- ✓ User acceptance testing across three deal stages
"Our reps used to spend the first hour of every day just doing CRM admin. Now the AI scores the leads and drafts the emails overnight — they walk in and just hit send."
Ready to Supercharge Your Sales Team?
Let's discuss how AI-powered lead scoring and email drafting can help your team close more deals with less admin.